mother or motherland: can a government have an impact on ...ftp.iza.org/dp4954.pdf · mother or...
TRANSCRIPT
DI
SC
US
SI
ON
P
AP
ER
S
ER
IE
S
Forschungsinstitut zur Zukunft der ArbeitInstitute for the Study of Labor
Mother or Motherland: Can a Government Have an Impact on Educational Attainment of the Population? Preliminary Evidence from India
IZA DP No. 4954
May 2010
Sumon Kumar BhaumikManisha Chakrabarty
Mother or Motherland:
Can a Government Have an Impact on Educational Attainment of the Population?
Preliminary Evidence from India
Sumon Kumar Bhaumik Aston Business School
and IZA
Manisha Chakrabarty Indian Institute of Management
Discussion Paper No. 4954 May 2010
IZA
P.O. Box 7240 53072 Bonn
Germany
Phone: +49-228-3894-0 Fax: +49-228-3894-180
E-mail: [email protected]
Any opinions expressed here are those of the author(s) and not those of IZA. Research published in this series may include views on policy, but the institute itself takes no institutional policy positions. The Institute for the Study of Labor (IZA) in Bonn is a local and virtual international research center and a place of communication between science, politics and business. IZA is an independent nonprofit organization supported by Deutsche Post Foundation. The center is associated with the University of Bonn and offers a stimulating research environment through its international network, workshops and conferences, data service, project support, research visits and doctoral program. IZA engages in (i) original and internationally competitive research in all fields of labor economics, (ii) development of policy concepts, and (iii) dissemination of research results and concepts to the interested public. IZA Discussion Papers often represent preliminary work and are circulated to encourage discussion. Citation of such a paper should account for its provisional character. A revised version may be available directly from the author.
IZA Discussion Paper No. 4954 May 2010
ABSTRACT
Mother or Motherland: Can a Government Have an Impact on Educational
Attainment of the Population? Preliminary Evidence from India* In this paper, using data from the 61st round of the (Indian) National Sample Survey, we examine the relative impacts of personal-household and state-level characteristics (including government policy) on the likelihood of transition from one educational level to the next. Our analysis suggests that the most important factors driving these transition likelihoods are personal and household characteristics like gender and education of household heads. However, state-level characteristics and government policies have a significant impact on these transition likelihoods as well, especially for transitions from the lowest levels of education to somewhat higher levels. The odds of making the transition to higher education, especially tertiary education, are systematically lower for women than for men, for individuals in rural areas than those in urban areas, and for Muslims than for Hindus. An important conclusion of our analysis is that there is significant scope for government policy to address educational gaps between various demographic and other groups in the country. JEL Classification: I21, I28 Keywords: educational attainment, likelihood of transition, government policy Corresponding author: Manisha Chakrabarty Economics Group Indian Institute of Management Joka, Diamond Harbour Road Calcutta 700 104 India E-mail: [email protected]
* The authors would like to thank Indian Institute of Management, Calcutta for financial support for the research, the National Council of Applied Economic Research for some of the data used in the analysis, and Shuvro Mondal for excellent research assistance. They remain responsible for all remaining errors.
1. Introduction
Education, which is an investment in human capital, plays a critical role in shaping a country’s
economic future. To begin with, there is a broad consensus about the positive impact of the stock of
human capital on a country’s growth rate (Barro, 1991; Mankiw, Romer and Weil, 1992). It is also
generally accepted that there are positive and significant returns to education, and that differences in
education can explain a significant proportion of earnings differences between various socio-
economic groups (Bhaumik and Chakrabarty, 2009a, 2009b) and indeed between labourers in
different countries (Gregorio and Lee, 2002; Bargain et al., 2009). There is also some evidence to
suggest that (unsurprisingly) the returns to investment in education are higher for people from the
more disadvantaged socio-economic classes (Krueger and Lindahl, 2001). Provision of education,
therefore, remains a key pillar of policymaking.
However, policies that emphasise removal of supply side constraints for spread of education
do not necessarily succeed in ensuring in meeting their stated objectives. Drop-out rates are high in
most developing countries. Even in many developed countries, a relatively small proportion of the
population receive university education. Formulating policies that aim to deliver more than literacy –
skills that require completion of high school or even university education – to a significant proportion
of the population, therefore, requires an understanding of factors that influence individual choice of
education levels. The aim of this paper is to make a contribution towards that policy discussion by
examining the impact of individual and household characteristics as well as government policy on
educational attainment in India.
It is well understood that educational attainment of individuals depends significantly on
personal characteristics and family backgrounds (Lave, Cole and Sharp, 1981; Teachman, 1987;
Lauer, 2003). In particular, it depends on the educational background of the individual’s parents and
the on the permanent income of the household (Tansel, 1997, 2002). Other studies have emphasised
the importance of mother’s education, and factors like nutrition that are related to a household’s
economic status (Zhao and Glewwe, In press). There is some evidence to suggest that the importance
of family background on educational attainment of individuals in developing countries is fairly stable
over time (Smith and Cheung, 1986). Religion and ethnicity can also play an important role in
determining an individual’s educational attainment, with some people from some religious and ethnic
backgrounds having a greater statistical likelihood of higher educational attainment than others
(Sander, 2009, In press). In part, this could be on account of inter-group differences in the impact of
parental education on educational attainment (Gang and Zimmerman, 2000). Educational attainment
is also affected by factors that affect an individual’s demand for education, as students respond to
economic incentives in making education choices (Wilson, 2001).
The evidence about the impact of government policy on educational attainment is much more
ambiguous, especially in the context of developing countries. There is evidence to suggest that
government policies, in part in the form of greater educational spending, can have a positive impact
on educational attainment of a population (King and Lillard, 1987; Gupta, Verhoeven and Tiongson,
2002). But the impact of government spending varies across countries (Gupta and Verhoeven, 2001).
Further, analyses using individual level data demonstrate that once factors like ability (which, in turn,
may be influenced by family background) are controlled, school characteristics like teacher-student
ratio that can be influenced by government policy no longer has any impact on educational attainment
(Dearden, Ferri and Meghir, 2002).
We examine the relative importance of family background (encompassing both individual and
household characteristics) and government policy on educational attainment in India. Specifically, we
use the 61st round of the National Sample Survey (NSS) data for 2005 to examine the impact of these
individual, household and regional characteristics influenced by government policies on the likelihood
of transition across educational levels (primary, middle, higher secondary and tertiary). In light of the
evidence about significant differences in the educational attainments of Hindus and Muslims in India
(Bhaumik and Chakrabarty, 2009a, 2009b), we separately estimate the impact of these variables on
the educational attainments of these two religious groups. Our results suggest that state-level
characteristics like per capita GDP and structure of a state’s economy do influence the likelihood of
transition from any level of education to the next (or higher) level. Government policies (captured by
the share of a government’s expenditure spent on education) matter as well. However, while public
spending on education has a positive impact on transition probabilities for lower levels of education,
they have a (counterintuitive) negative impact on the likelihood of transition from high school to
tertiary education.
The rest of the paper is structured as follows: In Section 2, we discuss the data and highlight
some interesting patterns. The econometric methodology is discussed in Section 3. In Section 4, we
report and discuss the implications of our regression results. Finally, Section 5 concludes.
2. Data
For our analysis, we use individual level data from the 61st round of the NSS. We concentrate on
individuals in the 25-30 age group. The lower limit for age is chosen on the basis of the reasonable
assumption that, with very few exceptions, an individual takes all her decisions about education (e.g.,
whether or not to enrol in a college or university) by the age of 25. The upper age limit is influenced
by the availability of data. As we shall see later, we argue that an individual’s decision to move from
the kth
education level to the (k+1)th
education level is influenced by the economic conditions
prevailing at the time at which the decision is taken. For example, the decision to enrol in a middle
school after the completion of primary education is made at the age of 14, such that for an individual
who was 30 year old in 2005, the year in which that decision was made was in 1989. We were able to
obtain appropriate data on economic conditions prevailing in different states in India going back to the
late eighties, particularly data on detailed break-up of state government’s budget, and this, in turn,
determined the upper limit of the age cohort for our analysis. We do not, however, consider this data
limitation to be a disadvantage. On account of these limits, all the individuals in our sample made
their educational decisions in the era of economic liberalisation in India which started in the mid
eighties (Rodrik and Subramanian, 2004), thereby making our analysis relevant in the current context
of a liberalised economy.
We concentrate on individuals from 12 states that account for 87% percent of the country’s
population and over 85 percent of its GDP. We leave the North Eastern states and Jammu and
Kashmir out of our sample because political uncertainties and insurgencies in these states may have
impacted decisions about educational attainment in ways that would be difficult to model empirically.
Further, we combine states like Jharkhand and Bihar that were a combined political entity in the early
nineties. This aggregation was necessitated by the fact that individual and household level data from
2005 had to be matched with state level data from the late eighties and the nineties, when these states
were unified political entities, which implies total number of states as 15.
Figure 1
Our final sample has 14,332 observations, of which 12,283 are Hindus and 2,049 are
Muslims. In keeping with earlier literature on India that also used NSS data (Bhaumik and
Chakrabarty, 2009a, 2009b), we distinguish between four levels of educational attainment: primary,
middle school, higher secondary (i.e., high school graduation), and tertiary which includes graduate
and above. The distribution of the Hindus and Muslims (and the overall sample) across the four
educational levels are reported in Figure 1. As highlighted in previous studies, while the overall
distribution is skewed in favour of lower levels of education, with primary and middle school
education accounting for 60.59 percent of the sample, the distribution is more skewed for the Muslims
(73.94 percent) than for the Hindus (58.37 percent). The advantage of the Hindus is particularly high
for tertiary education; 16.63 percent of the Hindu individuals in our sample have tertiary education,
double the proportion of the Muslims (8.2 percent).
Since the aim of this analysis is to examine the relative importance of family background and
government policies in determining educational attainment, it would be important to have a
significant variation in the characteristics of the states included in the sample. We distinguish between
two sets of government policies, the “flow” and the “stock”. We take into consideration the
contemporaneous education policy of the government as captured by the share of education in
government expenditure. We also take into consideration the economic status of each state – as
reflected by its development status (per capita state real GDP), the literacy rate of 1981, and
dependence on agriculture (contribution of agriculture to state GDP) – that is an outcome of policies
pursued over a number of years. While these factors affected the decisions taken by the individuals in
our sample in the late eighties and nineties, in order to provide a snap shot of inter-state variations in
these factors, in Figure 2 we report the average values of the underlying variables for the 1985-1998
period, whose relevance would be evident shortly. The diversity of the states is apparent from Figure
2.
Figure 2
If government policies, whether directly related to education or affecting behaviour of
economic agents by way of environmental factors like the level of development, do have a significant
impact on educational attainment, we should expect a significant variation in the aggregate levels of
educational attainment across the states. In Figure 3, we report the differences in the educational
attainment of individuals in our sample across the 12 states. It can be seen that while in each state the
share of primary and middle school education exceeds the share of higher/tertiary education by a
substantial margin, there are nevertheless significant variations across the states.
Figure 3
Education attainment not only varies across states, but location within states – broadly
divided into urban and rural locations – also matter significantly. In Figures 4a and 4b, we report the
distribution in individuals with middle school education and tertiary education across states, divided
into rural and urban locations. It can be seen that for lower levels of educational attainment, urban
locations do not have a significant advantage over rural locations in any of the states. However, for
higher/tertiary education, the advantage of urban locations is significant.
The above discussion suggests that there is considerable variation in educational attainment
across the Indian states, and there are also considerable variations in educational policies of
governments and other local conditions (that are affected by the “stock” of government policies) that
can affect an individual’s demand for education. Taken together, there is perhaps prima facie evidence
that government policies, whether about education itself or about the economics of the states in
general, might have an impact on educational attainment. However, there is also evidence to suggest
that factors like religion might influence an individual’s educational attainment, and we have not yet
looked at factors like parental education. Hence, at this stage, it is not possible to make a conjecture
about the relative importance of family background and government policies-local economic
conditions in determining educational attainment. We examine this more rigorously in the rest of this
paper.
Figure 4a Figure 4b
3. Methodology and specification
In contrast to the section of the literature that uses ordered probit to model the educational attainment
of individuals, we view progression through educational levels as a sequential process in which
attaining each level of education is conditional on not exiting the process after completing the
previous level of education. This view is consistent with the observation that children can (and indeed
do) drop out of schools after completing some years of education, and that not all high school
graduates continue into tertiary education. While there is a well-defined order in education – tertiary
education is higher than high school education, for example – the sequence and the risk of not making
the transition from one level of education to the next cannot be ignored.
Following Buis (2009), therefore, we model educational attainment in India using a sequential
logit model. As mentioned above, in light of the data availability and also the past literature on the
impact of education on earnings in India, we construct four levels of educational attainment, namely,
primary, middle school, higher secondary (or high school graduation) and tertiary. Given these levels
of education, we construct a sequence structure that is depicted in Figure 5.
Figure 5
Primaryeducation
Exit
Middle school
Exit
Highersecondary
Exit
Tertiary
p1
1 – p1
p2
p3
1 – p2
1 – p3
After completing any level of education k, an individual i has the option to continue to the
next level of education with probability pki or exit with probability (1 – pki). The use of the sequential
logit model yields estimates of these transitional probabilities pki that are given by
n
nkn
m
mkmk
n
nknm
m
kmk
ki
zx
zx
p
exp1
exp
ˆ , if pk-1,i = 1
We model the transition probability as a function of m individual and household characteristics (x)
and n other variables that capture the government’s educational policy of the individual’s state of
residence and the economic environment in the state in general. Starting from an educational level l,
an individual’s probability of reaching a higher education level L is, therefore, given by
L
lk
kp .
Table 1
Variable Measurement
Dependent variable Education = 1: primary or below-primary (up to class 5);
Education = 2: middle (up to class 8);
Education = 3: (higher) secondary (up to class 12);
Education = 4: undergraduate and above
Personal and household characteristics
Gender
Dummy variable = 1 for female (7588 males and 6744
females)
Household per capita consumption Mean = INR 687.67
Education of household head
Categorical variable with 1 indicating illiteracy and 13
indicating postgraduate education
Location Dummy variable = 1 for rural
Government policy and economic environment
Per capita state GDP
Data obtained from the Reserve Bank of India, and
measured in INR in 1993-1994 prices
Agriculture % of state GDP
Data was provided by the National Council of Applied
Economic Research
Literacy rate in the state State-level literacy rate in 1981
Education % of state govt. expenditure Data obtained from state government budget documents
The choice and measurement of household characteristics and the policy-environmental
variables are highlighted in Table 1. Our measures of personal and household characteristics are easily
understood. These measures are contemporaneous, i.e., of 2005. Some of these characteristics (e.g.,
gender) are invariant over time. Others like a household’s socio-economic status, measured by per
capita household consumption can, in principle, change over time. But it is possible to make the
reasonable assumption that in a developing country like India current socio-economic status is
strongly correlated with past socio-economic status such that, at the very least, the relative positions
of households in the distribution do not change substantially over time.
The variables capturing state-level characteristics and government policies, however, do not
have contemporaneous measure. Consider, for example, an individual who is 25 in 2005. If he took
the decision to make (or not make) the transition from middle school to (higher) secondary education
at the age of 14, then her decision would have been influenced by state-level characteristics and
government policy at that point in time, i.e., in 1994. For the same transition, the relevant year for an
individual who is 30 years old in 2005 is 1989. It is easy to see how (with one exception) the values
for the state-level variables were chosen for the analysis. Given the age range of 25-30 for our sample
of individuals, and given that the transitions range from “primary to middle school” to “higher
secondary to tertiary”, the values of the state-level characteristics and proxies for government policies
were chosen from the 1985-1998 period. The only exception to this is the literacy rate at the state
level. For this variable, we use an initial value for all states and all individuals, namely, the state-level
literacy rate in 1981. The rationale for the choice of 1981 as the initial year is that in that year all the
individuals in our sample were below the age of 5, which is roughly the age at which children in India
are introduced to formal education.
4. Regression results and discussion
The regression estimates are reported in Tables 2 (for Hindus) and 3 (for Muslims). Each of these
tables has three panels. Panel A reports the coefficient estimates for logit regressions for moving from
the primary education to any of the higher levels of education. Panel B reports the coefficients for
moving from middle school to higher secondary or tertiary education. Finally, Panel C reports the
estimated coefficients for moving from higher secondary to tertiary education. In both tables, for each
of these panels, most of the estimated coefficients are significant at the 5 percent or 1 percent level.
The likelihood ratio chi-square statistics for the regression models are also significant at the 1 percent
level. Hence, we are fairly confident that our specification explains variations in the educational
attainment in the data reasonably well, for both the Hindu and Muslim sub-samples.
Table 2
Transition 1
Panel A
Transition 2
Panel B
Transition 3
Panel C
Personal and household characteristics
Gender (female = 1)
- 1.034 ***
(0.051)
- 0.237 ***
(0 .057)
- 0.118 *
(0.064)
Household per capita consumption 0.001 *** (0.0001)
0.001*** (0.0001)
0.001 *** (0.00001)
Education of household head
0.300 ***
(0.009)
0.177 ***
( 0.009)
0.139 ***
(0.010)
Government policy and economic environment
Per capita state GDP
0.001 ***
(0.00003)
0.001 ***
(0.00003)
- 0.000004
(0.00003)
Agriculture % of state GDP
0.060 ***
(0.005)
0.070 ***
(0.006)
0.018 **
(0.007)
Literacy rate in the state
- 0.049 ***
(0.003)
- 0.050 ***
(0.003)
0.001
(0.0036)
Education % of state govt. expenditure
0.167 ***
(0.010)
0.078 ***
(0.012)
- 0.032 ***
(0.012)
Location
Rural household
- 0.377 ***
(0.059)
- 0.201 ***
(0.060)
- 0.444 ***
(0.064)
Regression statistics Log likelihood =-11913.753
LR chi-square = 9032.81
Sample size = 12283
Note: (1) Transition 1 is from primary to middle school or higher; Transition 2 is from middle school
to higher secondary or tertiary; Transition 3 is from higher secondary to tertiary. (2) Values within parentheses are standard errors. (3) ***, ** and * indicate significance at the 1, 5 and 10 percent
levels, respectively.
The coefficient estimates for the Hindu individuals (Table 2) suggest that the transition to a
higher level of education is affected both by household characteristics (or family background) and by
government policy and the economic environment prevailing in the state at the time of the relevant
decision. Both the educational attainment of the household head and the socio-economic status of the
household (as reflected in the per capita consumption of the household) have a significant and positive
impact on the likelihood of transition at each level of an individual’s educational attainment. Though
the effect household per capita consumption expenditure remains the same across three stages of
transitions, the impact of head education is more profound in the first stage of hurdle. In most cases,
being a woman reduces the likelihood of transition to the next level of educational attainment.
However, once an individual already attains higher secondary level of education, being a woman
though decreases the likelihood of transition to tertiary education, but the impact is only marginally
significant. This suggests that Hindu women in India generally tend to drop out of education early in
life.
Government policy and the economic environment have significant impact on the likelihood
of transition as well. Both government expenditure on education (as a percentage of total expenditure)
and the level of development in the state (as captured by per capita state GDP) have positive impact
on the likelihood of transition to middle school and higher secondary levels of education. However,
the level of development does not influence transition to tertiary education, while government
expenditure has a negative impact on the likelihood of this transition. In an equally counterintuitive
manner, state level literacy rate has a negative impact on the likelihood of transition to middle school
and higher secondary levels of transition. It is, however, not surprising that state-level literacy rate has
no impact on the likelihood of transition to tertiary education; the regional educational environment is
more likely to affect decisions to enrol in school, but perhaps not so much progression to tertiary
education. Interestingly, the likelihood of transition increases with the contribution of agriculture to
the state’s GDP, suggesting that education might be an instrument to signal capability and thereby
increase employability, and is particularly important in states where the spread of industries and the
services sector is low. Finally, unsurprisingly, residence in rural areas has a significant negative
impact on the likelihood of transition at all levels of educational attainment.
The regression results for Muslims (Table 3) are similar in most respects, but there are also
some differences. Once again, education level of the household head and the socio-economic status of
the household (captured by per capita consumption) have positive and significant impact on an
individual’s transition likelihood at each level of educational attainment. For Muslim individuals the
impact of household head education plays less dominant role, particularly at lower level of transitions,
reflected by the magnitude of the coefficient estimates. Being a woman reduces the likelihood of
transition from primary to middle school or higher levels of education, but has no or marginally
significant impact on the transition likelihood for higher levels of education.
Table 3
Transition 1
Panel A
Transition 2
Panel B
Transition 3
Panel C
Personal and household characteristics
Gender (female = 1)
- 0.780 ***
(0.125)
0.080
(0.144)
- 0.398 *
(0.213)
Household per capita consumption
0.002 ***
(0.0003)
0.001 ***
(0.0002)
0.001 ***
(0.0002)
Education of household head 0.236 *** (0.021)
0.150 *** (0.023)
0.139 *** (0.034)
Government policy and economic environment
Per capita state GDP
0.001 ***
(0.00001)
0.001 ***
(0.0001)
- 0.0003 ***
(0.0001)
Agriculture % of state GDP 0.118 *** (0.014)
0.071 *** 0.018
- 0.029 (0.026)
Literacy rate in the state
- 0.018 ***
(0.006)
- 0.041 ***
(0.007)
0.003
(0.010)
Education % of state govt. expenditure 0.221 *** (0.027)
0.079 *** (0.031)
- 0.099 *** (0.038)
Location
Rural household
0.247 *
(0.135)
0.082
(0.161)
- 0.196
(0.243)
Regression statistics Log likelihood = -1744.577 LR chi-square = 1477.01
Sample size = 2049
Note: (1) Transition 1 is from primary to middle school or higher; Transition 2 is from middle school to higher secondary or tertiary; Transition 3 is from higher secondary to tertiary. (2) Values within
parentheses are standard errors. (3) ***, ** and * indicate significance at the 1, 5 and 10 percent
levels, respectively.
Government policies and state level economic environment influence transition likelihoods as
well. Transition likelihoods for moving up from primary and middle school levels increase with the
per capita state GDP and with the share of education in the overall expenditure of the state
government. The impact of education expenditure is more prominent for the transition from primary
to higher secondary education for Muslim individuals than their Hindu counterpart. However, these
are negatively correlated with the likelihood of transition from higher secondary to tertiary education.
As with her Hindu counterpart, a Muslim individual’s transition likelihoods are inversely related to
the literacy rate of her state of residence. The share of agriculture does not affect the transition
likelihood from higher secondary to tertiary education. Unlike Hindu individual, the sector of
residence does not affect the transition likelihood for all levels of education.
Table 4
Hindu Muslim
Male Female Male Female
Urban Rural Urban Rural Urban Rural Urban Rural
Primary to middle
school or higher 0.93 0.79 0.82 0.57 0.70 0.68 0.50 0.48
Middle school to higher secondary
or tertiary
0.73 0.55 0.67 0.48 0.40 0.37 0.41 0.38
Higher secondary
to tertiary 0.45 0.25 0.39 0.21 0.31 0.24 0.19 0.14
Next, we compute the overall transition probabilities by religion, gender and (rural/urban)
location. They are reported in Table 4 and, in effect, are a reality check for our regression results. The
probabilities are consistent with our expectations. First, probability for transition is higher at lower
levels of education attainment than at higher levels. Even in the best of cases – for a Hindu male
resident in an urban area – the probability of transition from higher secondary to tertiary education is
0.45, less than half the transition probability from primary to a higher level of education. The odds
worsen even more rapidly for Muslims, women and residents of rural areas. Second, transition
probabilities are uniformly lower for females and members of rural households. This is evident from a
cursory comparison of the “urban” and “rural” columns for any given religious group and gender, and
the “male” and “female” columns of any given religious group and location. Finally, transition
probabilities are also uniformly lower for Muslims relative to their Hindu counterparts. Importantly,
while this is true for both men and women, the difference is starker for women than for men. For
example, for Hindu women in urban areas, the transition probability from primary to a higher level of
education (0.82) is more than 60 percent higher than the corresponding probability for an urban
Muslim woman (0.50). The extent of this gap is even greater (100 percent) for the transition
probability from higher secondary to tertiary education; 0.39 for the urban Hindu woman and 0.19 for
her Muslim counterpart.
Figure 6a
0.2
.4.6
.80
.2.4
.6.8
0.2
.4.6
.8
Rajasthan Assam West Bengal Gujrat
Maharashtra Andhra Pradesh Karnataka kerala
Tamil Nadu Madhya Pradesh Bihar Uttar Pradesh
Effect of Head education for Hindu Effect of head education for Muslim
Effect of head educaion on the expected highest level of education of urban female population with respective average values of head education
Figure 6b
0.2
.4.6
.80
.2.4
.6.8
0.2
.4.6
.8
Rajasthan Assam West Bengal Gujrat
Maharashtra Andhra Pradesh Karnataka kerala
Tamil Nadu Madhya Pradesh Bihar Uttar Pradesh
Effect of Head education for Hindu Individuals Effect of Head education for Muslim
Effect of household head education on the expected highest level of education for urban female population with average HIndu head's education
Finally, we revisit the question as to whether government policy has a role to play in
enhancing educational attainment, or whether much of it is determined by family background or
household characteristics. We have already seen from the regression estimates that government
policies – whether contemporaneous education policy or cumulative impact of economic policy
reflected in the level of development of the state – have at least as much impact on educational
attainment as household characteristics like education of household head and per capita consumption.
In light of our discussion about the differences in the transition probabilities of Muslim women
relative to their Hindu counterparts, we now focus on the importance of a key household characteristic
– education of the household head – which is believed to have a very significant influence on the
educational attainment of the household members. In Figure 6a, for each state, we report the impact of
the household head’s education on educational attainment of women, at the average education levels
of heads of Hindu and Muslim households. In Figure 6b, we recomputed this impact, after endowing
heads of Muslim households with the average education level of their Hindu counterparts. We can see
that while this bridges the gap between the educational attainment of Hindu and Muslim women, a
large part of the gap remains open. In other words, household characteristics in general and the
family’s educational background in particular do not explain the lion’s share of the inter-personal
variation in educational attainment (nor the difference in educational attainment of Hindus and
Muslims), leaving scope for appropriate government policy (whether targeted directly at education or
at the economic environment in general) to make an impact.
5. Conclusion
Education policies of governments should ideally take into account not just supply side failures but
also individual, household and state-level characteristics that might influence an individual’s decision
to continue with formal education, Mindful of this proposition, in this paper, we examine the relative
impacts of personal-household and state-level characteristics (including government policy) on the
likelihood of transition from one educational level to the next. We undertake the analysis separately
for Hindus and Muslims. Our analysis suggests that the most important factors driving these transition
likelihoods are personal and household characteristics like gender and education of household heads.
However, state-level characteristics and government policies have a significant impact on these
transition likelihoods as well, especially for transitions from the lowest levels of education to
somewhat higher levels. The odds of making the transition to higher education, especially tertiary
education, are systematically lower for women than for men, for individuals in rural areas than those
in urban areas, and for Muslims than for Hindus. These results are consistent with the existing
literature on gender gaps and gaps between Hindus and Muslims with respect to educational
attainment. An important conclusion of our analysis is that there is significant scope for government
policy to address educational gaps between various demographic and other groups in the country.
References
Barro, R.J. (1991). Economic growth in a cross-section of countries, Quarterly Journal of Economics,
106(2): 407-443.
Bhaumik, S.K. and Chakrabarty, M. (2009a). Is education the panacea for economic deprivation of
Muslims? Evidence from wage earners in India, 1987-2004. Journal of Asian Economics ,
20(2): 137-149.
Bhaumik, S.K. and Chakrabarty, M. (2009b). Earnings inequality in India: Has the rise of caste and
religion based politics in India had an impact? In: A. Shariff and R. Basant (Eds.) Handbook of Muslims in India, Oxford University Press, New Delhi, pp. 235-253.
Bargain, O., Bhaumik, S. K, Chakrabarty, M. and Zhong Zhao, (2009), Earnings Differences
Between Chinese And Indian Wage Earners, 1987-2004, Review of Income and Wealth, Blackwell Publishing, vol. 55(s1), pages 562-587, 07.
Buis, Maarten L.(2009), Not all transitions are equal: The relationship between inequality of educational opportunities and inequality of educational outcomes. in: Maarten L. Buis,
Inequality of Educational Outcome and Inequality of Educational Opportunity in the
Netherlands during the 20th Century, Chapter 6, pp. 105-136.
De Gregorio, J. and Lee, J-W. (2002). Education and income inequality: New evidence from cross-
country data, Review of Income and Wealth, 48: 395-416.
Dearden, L., Ferri, J. and Meghir, C. (2002). The effect of school quality on educational attainment
and wages, Review of Economics and Statistics, 84(1): 1-20.
Gang, I.N. and Zimmerman, K.F. (2000). Is child like parent? Educational attainment and ethnic
origin, Journal of Human Resources, 35(3): 550-569.
Gupta, S. and Verhoeven, M. (2001). The efficiency of government expenditure: Experiences from
Africa, Journal of Policy Modeling, 23(4): 433-467.
Gupta, S., Verhoeven, M. and Tiongson, E.R. (2002), The effectiveness of government spending on education and health care in developing and transition economies, European Journal of
Political Economy, 18(4): 717-737.
Lauer, C. (2003). Family background, cohort and education: A French-German comparison based on
a multivariate ordered probit model of educational attainment, Labour Economics, 10(2): 231-
251.
Lave, C., Cole, M. and Sharp, D. (1981). Determinants of education achievement, Economics of
Education Review, 1(2): 253-262.
King, E.M. and Lillard, L.A. (1987). Education policy and schooling attainment in Malaysia and
Philippines, Economics of Education Review, 6(2): 167-181.
Krueger, A.B. and Lindahl, M. (2001). Education and growth: Why and for whom? Journal of
Economic Literature, 39: 1101-1136.
Mankiw, N.G., Romer, D. and Weil, D.N. (1992). A contribution to empirics of economic growth, Quarterly Journal of Economics, 107(2): 407-437.
Rodrik, D and Arvind Subramanian, (2004), From Hindu Growth to Productivity Surge: The Mystery
of the Indian Growth Transition, NBER Working Papers 10376, National Bureau of Economic Research, Inc.
Sander, W. (1992). The effects of ethnicity and religion on educational attainment, Economics of
Education Review, 11(2): 119-135.
Sander, W. (In press). Religious background and educational attainment: The effects of Buddhism,
Islam and Judaism, Economics of Education Review, doi:10.1016/j.econedurev.2009.08.004. Smith, H.L. and Cheung, P.P.L. (1986). Trends in the effect of family background on educational
attainment in the Philippines, American Journal of Sociology, 91(6): 1387-1408.
Tansel, A. (1997). Schooling attainment, parental education and gender in Cote d’Ivoire and Ghana,
Economic Development and Cultural Change, 45: 825-856.
Tansel, A. (2002). Determinants of school attainments of boys and girls in Turkey: Individual, household and community factors, Economics of Education Review, 21(5): 455-470.
Teachman, J.D. (1987). Family background, educational resources and educational attainment, American Sociological Review, 52(4): 548-557.
Wilson, K. (2001). The determinants of educational attainment: Modeling and estimating human capital model and the education production functions, Southern Economic Journal, 67(3):
518-551.
Zhao, M. And Glewwe, P. (In press). What determines basic school attainment in developing countries? Evidence from rural China, Economics of Education Review,
doi:10.1016/j.econedurev.2009.10.008.